GNSS Trajectory Anomaly Detection Using Similarity Comparison Methods for Pedestrian Navigation

• Main Authors: Pekka Peltola, Jialin Xiao, Terry Moore, Antonio R. Jiménez, Fernando Seco
• Format: Article
• Language: English
• Published: MDPI AG 2018-09-01
• Series: Sensors
• Subjects: similarity; GNSS trajectory; pedestrian dead reckoning; multipath; anomaly detection
• Online Access: http://www.mdpi.com/1424-8220/18/9/3165

The urban setting is a challenging environment for GNSS receivers. Multipath and other anomalies typically increase the positioning error of the receiver. Moreover, the error estimate of the position is often unreliable. In this study, we detect GNSS trajectory anomalies by using similarity comparison methods between a pedestrian dead reckoning trajectory, recorded using a foot-mounted inertial measurement unit, and the corresponding GNSS trajectory. During a normal walk, the foot-mounted inertial dead reckoning setup is trustworthy up to a few tens of meters. Thus, the differing GNSS trajectory can be detected using form similarity comparison methods. Of the eight tested methods, the Hausdorff distance (HD) and the accumulated distance difference (ADD) give slightly more consistent detection results compared to the rest.